> Energy/performance wise they are poison, every byte has to be parsed twice, once by the middleman and once by the terminal

I get that software bloat is a problem, but it's a little strange to hear a man whine about the overhead of a byte getting parsed twice as part of a terminal session when ten years ago desktop computers had around 8-9GB/sec memory bandwidth and single-thread performance measured in multiple billions of operations per second...

...when he refuses to split out highly dynamic functions (the news parsers) of Calibre into a plugin so that he doesn't have to version-bump Calibre every time a newspaper slightly changes their website and breaks the parser, forcing tens of thousands of users to update the entire package (and he doesn't implement a self-updater, much less one with binary diff capability.)